From: "Daniel Kral" <d.kral@proxmox.com>
To: "David Riley" <d.riley@proxmox.com>, <pve-devel@lists.proxmox.com>
Subject: Re: [PATCH pve-access-control v2 05/10] fix #7294: acl: pool: add SDN VNets as pool members
Date: Mon, 06 Jul 2026 13:56:15 +0200 [thread overview]
Message-ID: <DJRGWNICETBA.1ED2HNONXOCNG@proxmox.com> (raw)
In-Reply-To: <20260626131035.112374-6-d.riley@proxmox.com>
Generally, it would be good if this patch also introduces the
network_list entry in the repository's README, where the pool fields are
documented for a quick overview.
On Fri Jun 26, 2026 at 3:10 PM CEST, David Riley wrote:
> diff --git a/src/PVE/RPCEnvironment.pm b/src/PVE/RPCEnvironment.pm
> index 7591aa9..d9372a0 100644
> --- a/src/PVE/RPCEnvironment.pm
> +++ b/src/PVE/RPCEnvironment.pm
> @@ -53,6 +53,21 @@ my $compile_acl_path = sub {
> $data->{poolroles}->{"/storage/$storeid"}->{$role} = 1;
> }
> }
> +
> + for my $network_key (keys $d->{network}->%*) {
> + my ($type, @path) = split('/', $network_key);
> +
> + if ($type eq 'vnet') {
> + my ($zoneid, $vnetid, $vlan) = @path;
> +
> + my $acl_path = "/sdn/zones/$zoneid/$vnetid";
> + $acl_path = "$acl_path/$vlan" if defined($vlan);
nit: use the .= operator here
> +
> + for my $role (keys $pool_roles->%*) {
> + $data->{poolroles}->{$acl_path}->{$role} = 1;
> + }
> + }
> + }
> }
> }
>
> @@ -63,15 +78,30 @@ my $compile_acl_path = sub {
> # means the role is set
> my $roles = PVE::AccessControl::roles($cfg, $user, $path);
>
> + my $poolroles_path = $data->{poolroles}->{$path};
the factoring of $data->{poolroles}->{$path} should be done in a
separate commit beforehand so the changes done in this commit are easier
to understand and review.
> +
> + # Pool ACL paths are setup without propagation, therefore checking
> + # /sdn/zones/<zone>/<vnet>/<tag> fails even if the user has the
> + # permission for the base path /sdn/zones/<zone>/<vnet>.
> + # To allow this, the roles of the base VNet path are looked up if
> + # the exact tagged path is not found in the pool.
> + if (!defined($poolroles_path) && $path =~ m|^/sdn/zones/[^/]+/[^/]+/\d+$|) {
> + my @parts = split('/', $path);
> + my $base_vnet_path = join('/', @parts[0 .. 4]); # remove tag
> +
> + # Inherit the permissions of the base VNet path for this request
> + $poolroles_path = $data->{poolroles}->{$base_vnet_path};
> + }
> +
nit: it's probably better if this is moved before the declaration of
$roles so it's clear that it isn't depending on it/interacting with
it, but only the code below is
Besides that, I'm curious whether there might be an use case where this
might not be wanted? AFAIK, assigning a VM to a VNet does not include
the network as assigning a VM to a VNet VLAN tag would have, so the
propagation might be a bit off here...
It would be possible to introduce the syntax
vnet/<zoneid>/<vnetid>/*
to make the propagation to the individual vlan tag objects more
explicit, so this can still be an opt-in feature, and otherwise
vnet/<zoneid>/<vnetid>
only grants assigning the VNet to pool guests.
Though this is only an idea.
> # apply roles inherited from pools
> - if ($data->{poolroles}->{$path}) {
> + if ($poolroles_path) {
> # NoAccess must not be trumped by pool ACLs
> if (!defined($roles->{NoAccess})) {
> - if ($data->{poolroles}->{$path}->{NoAccess}) {
> + if ($poolroles_path->{NoAccess}) {
> # but pool ACL NoAccess trumps regular ACL
> $roles = { 'NoAccess' => 0 };
> } else {
> - foreach my $role (keys %{ $data->{poolroles}->{$path} }) {
> + for my $role (keys $poolroles_path->%*) {
> # only use role from pool ACL if regular ACL didn't already
> # set it, and never set propagation for pool-derived ACLs
> $roles->{$role} = 0 if !defined($roles->{$role});
> @@ -219,6 +249,8 @@ sub compute_api_permission {
> $res->{vms}->{$priv} = 1;
> } elsif ($priv =~ m/^Datastore\./) {
> $res->{storage}->{$priv} = 1;
> + } elsif ($priv =~ m/^SDN\./) {
> + $res->{sdn}->{$priv} = 1;
What is the intent here to add this?
I'm not too familiar with compute_api_permission(), but it seems like
it's only relevant for OpenID and TFA authentication?
> } elsif ($priv eq 'Permissions.Modify') {
> $res->{storage}->{$priv} = 1;
> $res->{vms}->{$priv} = 1;
> @@ -274,6 +306,17 @@ sub get_effective_permissions {
> foreach my $storeid (keys %{ $d->{storage} }) {
> $paths->{"/storage/$storeid"} = 1;
> }
> +
> + for my $network_key (keys $d->{network}->%*) {
> + my ($type, @path) = split('/', $network_key);
> +
> + if ($type eq 'vnet') {
> + my ($zoneid, $vnetid, $vlan) = @path;
> + my $vnet_path = "/sdn/zones/$zoneid/$vnetid";
> + $vnet_path .= "/$vlan" if defined($vlan);
> + $paths->{$vnet_path} = 1;
> + }
This might benefit from a warn/log_warn if there's a malformed network
pool entry or one with an unknown type... Though this might not get
shown properly to users as this is only called from an API endpoint?
> + }
> }
>
> my $perms = {};
> @@ -353,6 +396,25 @@ sub check_sdn_bridge {
> }
> }
>
> + # check access to VLANs via pools
> + for my $pool (keys $cfg->{pools}->%*) {
> + my $poolcfg = $cfg->{pools}->{$pool};
> + next if !$poolcfg->{network};
> +
> + for my $network_key (keys $poolcfg->{network}->%*) {
> + my ($type, @path) = split('/', $network_key);
> +
> + if (defined($type) && $type eq 'vnet') {
> + my ($zoneid, $vnetid, $vlan) = @path;
> +
> + if ($zoneid eq $zone && $vnetid eq $bridge && defined($vlan)) {
> + my $vlanpath = "/sdn/zones/$zoneid/$vnetid/$vlan";
> + return 1 if $self->check_any($username, $vlanpath, $privs, 1);
> + }
> + }
I think this should also have an else path, where we warn/log_warn that
there's a malformed network pool entry or one with an unknown type...
> + }
> + }
> +
> # repeat check, but fatal
> $self->check_any($username, $path, $privs, 0) if !$noerr;
>
> diff --git a/src/test/parser_writer.pl b/src/test/parser_writer.pl
> index ea2778e..a809e21 100755
> --- a/src/test/parser_writer.pl
> +++ b/src/test/parser_writer.pl
> @@ -238,24 +238,35 @@ my $default_cfg = {
> vms => {},
> storage => {},
> pools => {},
> + network => {},
> },
> test_pool_members => {
> 'id' => 'testpool',
> vms => { 123 => 1, 1234 => 1 },
> storage => { 'local' => 1, 'local-zfs' => 1 },
> pools => {},
> + network => { "vnet/zone1/vnet1" => 1, 'vnet/zone2/vnet2' => 1 },
> },
> test_pool_duplicate_vms => {
> 'id' => 'test_duplicate_vms',
> vms => {},
> storage => {},
> pools => {},
> + network => {},
> },
> test_pool_duplicate_storages => {
> 'id' => 'test_duplicate_storages',
> vms => {},
> storage => { 'local' => 1, 'local-zfs' => 1 },
> pools => {},
> + network => {},
> + },
> + test_pool_duplicate_networks => {
> + 'id' => 'test_duplicate_networks',
> + vms => {},
> + storage => {},
> + pools => {},
> + network => { "vnet/zone1/vnet1" => 1, "vnet/zone2/vnet2" => 1 },
> },
> acl_simple_user => {
> 'path' => '/',
> @@ -431,12 +442,15 @@ my $default_raw = {
> 'test_role_privs_invalid' => 'role:testrole:VM.Invalid,Datastore.Audit,VM.Allocate:',
> },
> pools => {
> - 'test_pool_empty' => 'pool:testpool::::',
> - 'test_pool_invalid' => 'pool:testpool::non-numeric:inval!d:',
> - 'test_pool_members' => 'pool:testpool::123,1234:local,local-zfs:',
> - 'test_pool_duplicate_vms' => 'pool:test_duplicate_vms::123,1234::',
> - 'test_pool_duplicate_vms_expected' => 'pool:test_duplicate_vms::::',
> - 'test_pool_duplicate_storages' => 'pool:test_duplicate_storages:::local,local-zfs:',
> + 'test_pool_empty' => 'pool:testpool:::::',
> + 'test_pool_invalid' => 'pool:testpool::non-numeric:inval!d::',
> + 'test_pool_members' =>
> + 'pool:testpool::123,1234:local,local-zfs:vnet/zone1/vnet1,vnet/zone2/vnet2:',
> + 'test_pool_duplicate_vms' => 'pool:test_duplicate_vms::123,1234:::',
> + 'test_pool_duplicate_vms_expected' => 'pool:test_duplicate_vms:::::',
> + 'test_pool_duplicate_storages' => 'pool:test_duplicate_storages:::local,local-zfs::',
> + 'test_pool_duplicate_networks' =>
> + 'pool:test_duplicate_networks::::vnet/zone1/vnet1,vnet/zone2/vnet2:',
> },
> acl => {
> 'acl_simple_user' => 'acl:1:/:test@pam:PVEVMAdmin:',
> @@ -696,6 +710,7 @@ my $tests = [
> $default_cfg->{test_pool_members},
> $default_cfg->{test_pool_duplicate_vms},
> $default_cfg->{test_pool_duplicate_storages},
> + $default_cfg->{test_pool_duplicate_networks},
> ],
> ),
> vms => default_pool_vms_with([$default_cfg->{test_pool_members}]),
> @@ -705,15 +720,37 @@ my $tests = [
> . "\n\n\n"
> . $default_raw->{pools}->{'test_pool_members'} . "\n"
> . $default_raw->{pools}->{'test_pool_duplicate_vms'} . "\n"
> - . $default_raw->{pools}->{'test_pool_duplicate_storages'} . "\n",
> + . $default_raw->{pools}->{'test_pool_duplicate_storages'} . "\n"
> + . $default_raw->{pools}->{'test_pool_duplicate_networks'} . "\n",
> expected_raw => ""
> . $default_raw->{users}->{'root@pam'}
> . "\n\n\n"
> + . $default_raw->{pools}->{'test_pool_duplicate_networks'} . "\n"
> . $default_raw->{pools}->{'test_pool_duplicate_storages'} . "\n"
> . $default_raw->{pools}->{'test_pool_duplicate_vms_expected'} . "\n"
> . $default_raw->{pools}->{'test_pool_members'}
> . "\n\n\n",
> },
> + {
> + name => "pool_format_backward_compatibility",
> + config => {
> + acl_root => default_acls(),
> + users => default_users(),
> + roles => default_roles(),
> + pools => default_pools_with([$default_cfg->{test_pool_empty}]),
> + },
> + raw => ""
> + . $default_raw->{users}->{'root@pam'}
> + . "\n\n\n"
> + # old format: 4 colons
> + . "pool:testpool::::\n\n\n",
> + expected_raw => ""
> + . $default_raw->{users}->{'root@pam'}
> + . "\n\n\n"
> + # new format 5: colons
> + . $default_raw->{pools}->{'test_pool_empty'}
> + . "\n\n\n",
> + },
nice!
> {
> name => "acl_simple_user",
> config => {
> @@ -1102,7 +1139,7 @@ my $tests = [
> . 'user:test@pam:0:0::::::' . "\n"
> . 'token:test@pam!test:0:0::' . "\n\n"
> . 'group:testgroup:::' . "\n\n"
> - . 'pool:testpool::::' . "\n\n"
> + . 'pool:testpool:::::' . "\n\n"
> . 'role:testrole::' . "\n\n",
> },
> ];
next prev parent reply other threads:[~2026-07-06 11:56 UTC|newest]
Thread overview: 23+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-06-26 13:10 [PATCH access-control/cluster/common/manager/network/proxmox-widget-toolkit/qemu-server v2 00/10] fix #7294: pool: add SDN VNets as pool members David Riley
2026-06-26 13:10 ` [PATCH pve-manager v2 01/10] ui: replace var with let to match style guide for variable declaration David Riley
2026-07-03 13:14 ` Daniel Kral
2026-06-26 13:10 ` [PATCH pve-manager v2 02/10] fix #7294: api: pool: add SDN VNets as pool members David Riley
2026-07-03 13:19 ` Daniel Kral
2026-06-26 13:10 ` [PATCH pve-manager v2 03/10] fix #7294: ui: " David Riley
2026-07-06 14:17 ` Daniel Kral
2026-06-26 13:10 ` [PATCH proxmox-widget-toolkit v2 04/10] fix #7294: css: theme: add opacity override for pool VNet icon David Riley
2026-07-03 13:30 ` Daniel Kral
2026-06-26 13:10 ` [PATCH pve-access-control v2 05/10] fix #7294: acl: pool: add SDN VNets as pool members David Riley
2026-07-03 14:35 ` Daniel Kral
2026-07-06 11:56 ` Daniel Kral [this message]
2026-06-26 13:10 ` [PATCH pve-network v2 06/10] fix #7294: sdn: register api formats for zones and vnets David Riley
2026-07-06 12:21 ` Daniel Kral
2026-06-26 13:10 ` [PATCH pve-network v2 07/10] fix #7294: sdn: vnet: update pool members on vnet migration and deletion David Riley
2026-07-06 12:29 ` Daniel Kral
2026-06-26 13:10 ` [PATCH pve-common v2 08/10] tools: add helpers for version comparison David Riley
2026-07-06 13:02 ` Daniel Kral
2026-06-26 13:10 ` [PATCH pve-cluster v2 09/10] fix #7294: cluster: helpers: add cluster-wide version assertion David Riley
2026-07-06 13:00 ` Daniel Kral
2026-06-26 13:10 ` [PATCH qemu-server v2 10/10] fix #7294: helpers: use cluster-wide version helper David Riley
2026-07-06 13:20 ` Daniel Kral
2026-07-06 14:07 ` [PATCH access-control/cluster/common/manager/network/proxmox-widget-toolkit/qemu-server v2 00/10] fix #7294: pool: add SDN VNets as pool members Daniel Kral
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=DJRGWNICETBA.1ED2HNONXOCNG@proxmox.com \
--to=d.kral@proxmox.com \
--cc=d.riley@proxmox.com \
--cc=pve-devel@lists.proxmox.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.